A study published in last month’s Archives of Internal Medicine has reignited the patient sat survey debate, correlating high patient satisfaction with several negative indicators, including higher patient death rates.
A study published in last month’s Archives of Internal Medicine has reignited the patient sat survey debate, correlating high patient satisfaction with several negative indicators, including higher patient death rates.
Over the years, many physicians have debated the value of patient satisfaction surveys. In 2010, I authored/co-authored several articles about inappropriate use and bias in patient satisfaction scores and how the invalid statistics adversely affected patient care and medical providers. Those articles made a lot of people upset and resulted in a rebuttal article in this publication by Press Ganey statisticians titled Patient Satisfaction Surveys are Here to Stay.
The fires of that conflict, having gone somewhat quiet, seem to have been rekindled thanks to a study published last month in Archives of Internal Medicine called “The Cost of Satisfaction.” The study reviewed patient satisfaction surveys from nearly 52,000 adults and correlated high patient satisfaction with several negative indicators, including higher patient death rates.
The study authors analyzed national Medical Expenditure Panel Survey (“MEPS”) data from 2000 to 2007 and then compared satisfaction scores with variables such as self-rated health status, health care utilization, health care expenditures, prescription drug expenditures, and future mortality. The authors even calculated the study data after excluding patients in poor health, theorizing that such patients would be more satisfied with their physicians, but would also be more likely to die from their diseases.
Results of the study showed that patients who had the highest satisfaction ratings spent 9% more on health care and prescription medications than did patients who had the lowest satisfaction ratings. In addition, the most satisfied patients had a 26% greater risk of death compared to least satisfied patients. When patients in poor health were excluded, the risk of death for these highly-satisfied “healthy” patients increased to 44% more than their least-satisfied counterparts.
In commentary accompanying the article, Dr. Brenda Sirovich suggested that discretionary testing is likely the cause of both the increased costs and the increased mortality in highly satisfied patients. Patient perceptions, even if medically inappropriate, drive testing and treatment. Antibiotics are harmful in patients with viral infections, yet a substantial subset of patients are not satisfied without an antibiotic prescription for their colds. Large studies show no link between PSA screening and either overall survival or prostate cancer survival. However, any patient whose life has been “saved” by a PSA screen is often quite satisfied. In both scenarios, there is no perceived negative effect from treatment. Patients will recover from their colds with or without antibiotics. Patients likely would not have died from their prostate cancer even if it was left untreated.
The concept of a doctor desiring to provide more testing with no perceivable negative effect is called a “positive feedback system” because the positive perceived benefit of treatment drives further treatment. The demand for antibiotics becomes greater with each successfully treated viral infection and the demand for “stronger” antibiotics grows with each perceived antibiotic failure. Repetitive PSA screens assure patients that no cancer is present even though the presence of cancer would likely not have changed the patient’s outcome. A positive feedback system may improve satisfaction, but, according to Dr. Sirovich, is also “unstable” and unable to “control its own growth or demise.”
Since its internet publication, the Archives article has been cited by several medical blogs and multiple newspapers, with such titles as “Do you like your doctor? It could be the death of you” and “A patient-pleasing doctor may not be the best.” Some commenters question whether it would be better for one’s health to have an angry doctor with a poor bedside manner. Other commenters note that the study conclusions cause a confusion between “correlation” and “causation.” Just because more patient deaths are correlated with high patient satisfaction does not mean that more patient deaths are caused by high patient satisfaction. But even if there is no causal relationship between death and satisfaction, the studies findings beg further exploration. At the very least, it would seem that if we emphasize high patient satisfaction over proper medical care, we are doing ourselves and our patients a disservice.
Does the study mean that patients who are more likely to die just happen to be more satisfied with their physicians? No. In fact, when the study excluded data from patients in poor health or with many medical problems, the rates of death for highly satisfied patients actually increased. Does the study mean that hospitals should shift their focus to hiring rude and insensitive physicians? Of course not. Despite some weaknesses in the data collection (such as the fact that costs and satisfaction were not measured in the exact same calendar years), the study suggests that we shouldn’t be so eager to equate “patient satisfaction” with “good medical outcomes.” Satisfaction is not always in a patient’s best interests.
The study findings have implications far beyond the relationship between physicians and their patients, though. From a public policy perspective, especially in the face of large budget deficits, the federal government is eventually going to realize that overemphasis on patient satisfaction increases the cost of medical care. By creating a monetary incentive to increase patient satisfaction, the government is not only increasing its expenses, but is also fostering some other variable that significantly increases patient death rates. Will the next headline be “HCAHPS kill?” From a legal standpoint, this study also demonstrates a growing conundrum for hospital administrators and hospital boards. If patient satisfaction statistics are biased and often not valid, if higher patient satisfaction rates increase hospital revenues, and if higher patient satisfaction scores also correlate with significantly higher patient death rates, will administrators soon be added to the list of defendants in medical malpractice/wrongful death lawsuits? How will administrators refute a plaintiff attorney’s allegation that they encouraged doctors to order discretionary testing that was detrimental to their patients’ interests in order to increase hospital profits?
Patient satisfaction scores may be “here to stay,” but as the Archives study shows, those who over-rely upon patient satisfaction results may do so at their own peril.
Dr. Sullivan practices emergency medicine in Illinois, is a clinical assistant professor of EM at the University of Illinois, and has a private law practice in Frankfort, IL
1 Comment
Dr. Sullivan,
Thank you for your insightful piece.
As an expert in customer experience, I too am concerned with the approach used to define patient satisfaction and the linkage of that to healthcare funding. However, I think there has to be a clear distinction in future studies.
When we (my company http://www.tonybodoh.com) have worked with medical practices (not hospitals) we have examined the differences in the commentary attached to patient reviews. Going beyond “satisfaction” we study what themes correlate with higher or lower satisfaction. And, some of what we’ve found could support the findings of this study: Some, but certainly not all, of the commentary of low rated reviews are a result of the patient not believing the diagnosis of the doctor. They therefore seek out another doctor to get the medical attention they were initially seeking with the first doctor. Is it possible that the patient satisfaction studies need to take a longitudinal approach to see if the dissatisfied patients sought treatment elsewhere and therefore their additional treatment is not included in hospital records? Is it possible that these dissatisfied patients had evidence that what they were not being evaluated sufficiently? These questions are still unanswered as far as I can tell.
Other reasons for low patient satisfaction relate to the time the doctor takes with the patient and with billing issues. Some doctors are actually penalized for suggesting additional testing because it is not covered by the patient’s insurance and the patients feel a “bait and switch” is happening.
On the other hand, when patients rate a doctor’s office or the doctor highly, the comments reflect the compassion of the doctor; their willingness to listen; their desire to care. But, it does not imply additional studies were done by the doctor.
For the research you commented on and the subsequent reporting of that research it seems there is a generalization being made that high scores leads to higher mortality. I believe the research needs to go deeper into the themes that patients mention as the reasons they are satisfied.
I believe a patient-centered focus is important and that satisfaction scores alone are not the right answer. In my experience companies that rely on just a satisfaction metric can make progressive improvements sometimes. Other companies using the same method fail to make noticeable improvements. The companies that consistently make the most improvement are those that learn to study the quantitative and the qualitative elements of the customer experience.
I would suggest the same is required for patient experience. There is no score that is sufficient. Scores only indicate that something may be right or wrong, not why it is right or wrong. That requires a different type of research that is not necessarily quantifiable.